Research On Limited Feedback Technology In Multiple-antennas Communication Systems

Multiple Input Multiple Output (MIMO) technology, which use antenna arrays rather than additional bandwidth for multiplying data rates, can overcome the problem of short radio spectrum resources well, and has been one of key issues in the next-generation wireless communication systems. When channel state information (CSI) is available at the transmitter, the performance of MIMO system can be further improved through transmitter pre-processing. In frequency division duplexing (FDD) system, channel state information at the transmitter (CSIT) can be obtained through a limited rate feedback link. The limited feedback technology, which researches on how to feedback and utilize CSIT, has been a highly active research area in multiple-antennas communication systems.In MIMO-OFDM systems and multiple-user MIMO systems, the feedback amount of CSI increases rapidly with number of antennas, number of subcarriers, and number of users. So how to reduce efficiently the feedback rate to improve the system throughput is an important problem to be solved. This dissertation aims to resolve the problem by summarizing simply the research status and basic principle of limited feedback technology and researching emphatically on the limited feedback technologies in temporally-correlated MIMO systems, MIMO-OFDM systems and multiple-user MIMO systems. The major innovations of this dissertation are described as below:For temporally-correlated MIMO systems, three limited feedback approaches are proposed. Firstly, a novel limited feedback approach based on neighborhood is proposed, including two schemes of both periodic feedback and auto-adaptive feedback. The new codebook for current frame consists of the codewords within the neighborhood of the codeword corresponding to previous frame, accordingly reducing a great deal of feedback bits. This approach can obtain almost similar performance with low complexity relative to existing feedback approach based on state transition probabilities.Then, another novel limited feedback approach based on tree-structure codebook is proposed, in which the channel vector of current frame is quantized into son-codewords of the codeword corresponding to previous frame. This approach can remarkably enhance the quantized precision of CSI and improve the BER performance under strong correlation. Finally, last novel limited feedback approach based on channel prediction is proposed. In the approach, CSI is fed back once every multiple data frames, and the optimized CSI codeword is selected with BER minimizing criterion using the auto-regressive prediction model to compensate the effect of feedback delay. This approach is the best one in proposed feedback approaches, and exceeds the existing limited feedback based on Markov chain with the same feedback rate.For MIMO-OFDM systems, three limited feedback approaches are proposed. Firstly, a novel clustering algorithm based on mean channel response is proposed, in which the mean CSI within cluster is quantized, and the computation complexity comparison with other clustering algorithms is presented. The simulation results show that the channel mean clustering algorithm can get good tradeoff between performance and complexity. Then, two feedback reduction approaches based on clustering are proposed. These feedback approaches can further reduce the feedback rate sufficiently exploiting the correlation adjacent clustering codewords including two schemes of both recursive feedback and trellis-based feedback. The former has slight performance loss with low complexity meanwhile the latter almost has little performance loss with large complexity. Finally, for temporally-correlated MIMO OFDM systems, a novel limited feedback approach based on two dimensions clustering in both time domain and frequency domain is proposed. A time-frequency resource block is clustering by extending the idea of clustering to time domain, and performance comparison of two dimensions clusterings with different optimization criterions is provided.For downlink multiuser MIMO systems, performances of several classical SDMA systems are compared, and both a novel scheduling algorithm and a novel feedback approach are proposed. Firstly, multiuser opportunistic beamforming system, PU2RC system, multiuser precoding system and their multiuser scheduling algorithms are introduced. Some important conclusions are educed by comparing and analyzing the simulation results of these limited feedback systems. Then, two improved greedy scheduling algorithms are proposed for ZF-SDMA system. These scheduling algorithms can insure that the most users are scheduled without the difficult problem of parameter setting relative to existing semi-orthogonal user selection (SUS) scheduling algorithm. The improved spatial correlation minimizing scheduling algorithm can achieve better sum rate than SUS scheduling algorithm under small users. Finally, a novel multiuser limited feedback approach based on graded quantization is proposed for ZF-SDMA system. In the proposed approach, multiuser scheduling uses coarse CSI quantization and multiuser precoding design uses precise CSI quantization. This approach can remarkably reduce the feedback amount or improve the sum rate performance with the same feedback rate compared with traditional ZF-SDMA system.